Management Metrics, A Primer

Post on 07-May-2022

2 views 0 download

Transcript of Management Metrics, A Primer

Management Metrics, A PrimerTim Adams

NASA Kennedy Space Center

2015 ASQ World Conference[Selected Slides]

Objectives

• Answer the “journalistic

questions” about

management metrics,

metrics for managing.

• Use these answers to

motivate the manager to

build and use metrics for

his/her organization.

1

Metric – a mathematical view

• In Mathematics, a metric

is an abstraction of the

notion of distance.

• Not all topologies* have a

metric space, and not all

metric spaces measure

distance the same way.

* Topology is concerned with properties of geometric figures that remain invariant when the

figures are deformed in any way that does not create new points or fuse existing points.

2

Management metric – flow diagram

3

GoalDesired

State

Current

State

MetricRisk

Analysis

Work

Process

Decision

Start

OK

Not OK

Not

Sure

• A management metric assigns a

measure to the performance gap.

• A mathematical function that

assigns a measure to the “gap”

between the output’s current

state (c) and the goal’s desired

state (d).

4

What is a management metric?

• Absolute Error

ΔAE = |d - c|.

• Relative Error

ΔRE = |(d - c)/d| where d ≠ 0.

• Relative Change

ΔRC = (c - d)/|d| where d ≠ 0.

5

Math for metrics – getting started

• Subjective

Produced by a state of mind

Particular to an individual

Example: “We are safe.”

• Objective

Material object, not mental concept

Actual existence

Not influenced by emotion or opinion

Example: “No accidents occurred.”

Management metrics – 2 types

“Subjective” metrics – example

7

“Objective” metrics – 3 types

1. Effectiveness (~ Quality)

How well results accomplished

the stated purpose.

2. Efficiency (~ Quantity)

How well resources were used

or consumed.

3. Appeal (~ Acceptance)

How well human preference

was satisfied.

8

Effectiveness metrics – 2 types

• Technical characteristics

Physical characteristics (e.g., size,

shape) and functional capability.

“On the drawing.”

• Operating characteristics

Non-physical characteristics being

operating behaviors and outcomes

(examples on next page).

“Inferred by the drawing.”

9

Operating characteristics – 9 types

10

Safety:

Freedom from

accident and loss

Usability:

Human interfaces

Supportability

and

Serviceability:

Service throughout

the planned life cycle

Reliability:

Likelihood of having

an uptime (failure-

free) state for a

stated duration or

load

Maintainability:

Likelihood of

returning to an

uptime state due to

maintenance or

repair

Availability:

Likelihood a

repairable item has

an uptime state;

A = f (R, M)

Producibility:

Ease and economy

of producing or

manufacturing

Affordability:

Total cost of

ownership and not

only system

acquisition cost

Disposability:

Disassembly and

disposal

(environmental

stewardship)

Where should metrics be used?

• Metrics are needed:

With imperfect systems.

When desired state is critical.

When management has a high

desire to obtain the desired state.

• Metrics are not needed

With perfect or low-risk systems.

11

Why do metrics?

• To quantitatively characterize

the performance gap and to

support the decision to…

Continue as is,

Make adjustments (changes), or

Obtain more information to

make a risk-informed decision.

Graphs can be misleading.

12

Great graph; why metric formulas?

13

Desired state = 2400 hr; Avg current state = 1300 hr)

Example What is the trend? (d = ?)

14

Trending without graphs

• All graphs used the same data!

• To test for a trend in discrete events

without graphing, use the Laplace

Test, a test statistic.

15

)12/(1]2/)/)t[(( i nTTnzn

i

ti is the time from the start time to the time of the ith event.

n is the number of events.

T is the time from the start time to the end time of the observation period.

Graphs instead of metric formulas?

• Sometimes “a picture is

worth a 1000 words” –

and sometimes it can be

misleading or confusing.

• When there is sufficient

amount of data, do both:

Plot the data

Treat the data.

16

When are metrics collected?

• During the game (Formative)

In process; is inferential.

• End of the game (Summative)

End of process; is descriptive.

• Not at all (not in the game?)

“Not keeping score” still allows

activity to be reported, or

Use subjective (self-rating) metrics.

17

How many metrics are needed?

• Consider a dashboard,

a combination of

metrics, for your

organization.

• Performance is seldom

assessed by a single

metric. Consider …

18

Pick one: “I need you to be…”

1. Effective

Complete task “x” with no errors.

2. Efficient

Produce “y” units per hour, cost.

3. Appealing

Check your work, support last

minute changes, be team

oriented, be safe...

19

Example – aggregating unlike scales

20

Sink, D.S., Productivity Management: Planning, Measurement and Evaluation, Control and Improvement, 1985, pp. 198-202.

Method – aggregating unlike scales

• The “Objectives Matrix” method:

Combines multiple-unlike measures

based on any scale into one score.

Different weights can be assigned to

each performance area.

The resulting composite value can be

trended over time.

Called multicriteria performance /

productivity measurement technique.

21

Who should make the metrics?

• The manager!

• Why, because…

• Making (not buying) your

management metrics:

Provides “expert/information power”

Can be revealing…

22

One reason to do your own metrics

“What is most remarkable is that the

mere effort to measure a difficult-to-

measure construct can lead to a

much deeper understanding and

more effective management of that

dimension or asset.”

23

Source is Dean Spitzer, author of Transforming Performance

Measurement, AMACOM, 2007. (Balestracci, D., “Measurement

As a Framework for Strategy,” Qualitydigest.com, May 08, 2013).

Benchmarking: Do your metrics …

• Measure distance or length?

Distance is between two points.

For management, the two states are:

oWhat you want, and

oWhat you have.

Length is a measure from zero.

o It does not compare states.

o It only measures one state.

24

Benchmarking: Do your metrics …

• Focus on what should be

measured, and not what

can be measured?

• Objectively measure all areas of

organizational performance?

Effectiveness,

Efficiency, and

Human Appeal?

25

A management metric is not a…

• Statistic, a function of the sample data.

• Trend when it uses length (not distance).

• Figure of merit, aggregated quantities used

to characterize performance and options.

• Risk measure, Prob. of Failure x Impact.

• Any single count or measure or just lengths.

26

Why? Because a metric is a function

of two points (states) and not one.

27

Timothy C. Adams

Reliability & Risk Engineer

NASA Kennedy Space Center

Engineering & Technology Directorate

Technical Performance & Integration Division

Tim.Adams@NASA.gov

321-867-2267

Contact Information